❓ Rovo Q&A Highlights: Top Contributors & Issues❓

[#1]

Huge thanks to everyone helping keep the Rovo Community Q&A active and growing — your contributions are making a real difference.

Responding to Rovo questions isn’t easy — new product + evolving AI = constantly moving target! Your thoughtful replies, experiments, and shared discoveries are building confidence for new users and keeping momentum strong while the documentation and features continue to mature.

This Round’s Top Contributors

A big round of applause for this month’s top five contributors!

🥇 @Rebekka Heilmann _viadee_  — Pulled ahead at the last moment to claim the #1 spot!
🥈 @Tomislav Tobijas — The unsung hero who came out of nowhere to secure #2.
🥉 @Nikola Perisic — Neck-and-neck battle with Rebekka for a while; outstanding consistency for #3.
🏅 @Alex Gallien (Atlassian) — Thank you for jumping in with clarity and guidance.
🏅 @Hector Menchaca (Atlassian) — Your participation shows users that Atlassian is here, watching, and supporting.

You're helping shape what the Rovo community becomes — collaborative, curious, and full of practical solutions.

Common Questions & How to Save Everyone Time

To cut down on duplicate posts and help new users find answers faster, here are the biggest themes emerging right now:

Dev CLI Confusion

A noticeable number of users are mixing up the Rovo Q&A forum with questions meant for Atlassian Dev CLI, especially around Forge, app development, and command-line usage. If your question involves:

• CLI commands

• Forge app scaffolding

• Deployment errors

• Local dev tooling

…it belongs in the Rov Dev Community rather than Rovo Q&A. They also have a form to report feedback and bugs.

Asking Rovo to Do Things It Cannot Access... yet?

Rovo can’t act on or “see” fields unless Atlassian exposes them through the Teamwork Graph. This means:

• Custom fields may not be interpretable.

• Some Jira or Confluence metadata isn’t accessible yet.

• Rovo can reference or summarize what it can pull — but not manipulate hidden fields or system-level objects.

If Rovo says something “doesn’t exist,” it may not be a bug — it may simply be outside its access scope at this time.

Known Limitations & Current Bugs

A few known issues that keep popping up across threads:

• Agent Skills not executing even though instructions are correct (often fixed by re-selecting the skill in Rovo Studio).

• Search hallucinating missing content when connectors haven’t finished indexing.

• Slack loop responses when permissions aren’t set correctly in the Slack integration.

• Agents only summarizing part of a JQL list — usually tied to data access or pagination limits.

• Skills that require confirmation not prompting correctly in some orgs.

These are all currently acknowledged or being monitored. When in doubt, tag your post with the details — patterns help Atlassian investigate faster.

Skills vs. Actions Confusion

This one’s still common:

• Skills = what the agent can do

• Actions = what the user triggers

Since a Skill can sometimes appear as an Action, the terms overlap more than users expect — leading to misunderstandings about what agents should or shouldn’t do.

Indexing, Permissions & RBAC

Many “Rovo isn’t working” questions end up being:

• Missing Confluence/Jira permissions

• Spaces or projects not selected for indexing

• RBAC still in beta and not fully rolled out

• Agents unable to act because the underlying Skill doesn’t have the right access

If Rovo can’t see it, it can’t act on it — no matter how perfect the prompt is.

Quotas, Usage & Dashboards

Many questions right now relate to confusion about AI credits, indexed object limits, and usage visibility:

• AI credit consumption is tracked but not yet enforced, so many users aren’t seeing measurable limits today.

• Indexed object quotas exist (each connector-item counts) and are tied to your licensed user count, but over-usage billing is not active yet. 

• Usage dashboards (to monitor credits, indexed object counts, top users, etc.) are still rolling out — you may not have full visibility yet. 

• Some features consume zero credits today (e.g., standard search queries) while others (like deep research or agent invocation) consume defined credit units.

It’s becoming increasingly important for Atlassian to provide clear, official guidance on how these quotas will work, how usage will be measured, and what customers should expect.

For now, the best approach is to test in a sandbox first to understand your organization’s patterns and be ready once Atlassian publishes the full consumption model.

How to Keep the Momentum Going

• Upvote helpful answers to give contributors visibility.

• Accept answers once your issue is resolved — it keeps the forum clean and highlights correct solutions.

• Check existing threads before posting — you’d be surprised how many questions share the same root cause.

Every upvote, comment, and accepted answer helps strengthen the whole knowledge base — and keeps the leaderboard meaningful and fun.

From @Kelly Stocker (Atlassian) and me (Rovo Community Champion) — a heartfelt thank-you to everyone who continues showing up, sharing answers, thinking out loud, and helping new users navigate Rovo. 💙

You’re not just solving problems — you’re building the culture, credibility, and heartbeat of this community.

(This post marks the end of the Q's I've reviewed in this round)

[Rebekka Heilmann _viadee_]

Thank you for the shout out (and accepting my answers so I *could* climb to the top ;) )!

Always learning something new. 

Skills vs. Actions Confusion

This one’s still common:

• Skills = what the agent can do

• Actions = what the user triggers

Since a Skill can sometimes appear as an Action, the terms overlap more than users expect — leading to misunderstandings about what agents should or shouldn’t do.

-> Can an Agent Skill be executed without a human in the loop? I thought they simply renamed Actions to Skills in Studio (and added more capabilities).